Wednesday, August 20, 2008

Going back to Natural Intelligence for a change

Ever typed a phrase into the Google search box only to scroll down to the 15th page of the one million hits returned to find what you were looking for? If you haven’t been living under a rock these past few years you probably have.

While it is relieving for us egotistical human beings to find that Google might not be infallible after all, there’s even better news to be had. The antidote to Google’s lack of discretion is not a computer, but rather the good old, archaic human mind.

Even in this seemingly post-natural world, a search tab cannot quite rival the precision of a human brain in its ability to discern information. Which is exactly why the increasingly popular search engine, ‘Ask ChaCha’, is manned by real people. Under the guise of the now familiar search box, thousands of human beings offer answers to questions, typing away at their computers.

Some binary code has to be transmitted somewhere in the universe before you can find out how many majors Tiger Woods has won, only you no longer have to battle with a confounded computer program to do so. The search engine also allows the option of different media forms. For those of us not yet savvy enough to own a BlackBerry, there is a number to call or text for that pressing 64 thousand dollar question that pops up on the subway ride home. ChaCha comes back with the reply in seconds. Think the phone-a-friend lifeline on ‘Who wants to be a Millionaire.’ This is like having a thousand smart friends, all at once.

Chacha however, as NPR’s Mark Phillips learned, is not quite as objective as a search engine might be. Anything that involves a human being couldn’t possibly come devoid of feelings and emotion and judgment. So while it can give you the name of the restaurant that serves the best thin-crust pizza in your town, and decide who will win the Superbowl this season, it also has the audacity to proclaim that homosexuality is wrong!

But even the seemingly innocuous Google search box is subjective in its own way – it picks out answers biased by statistics, location, and search history. Now, what's wrong with adding a little human subjectivity to that mix?

“Artificial artificial intelligence” for tasks that the human mind can perform better than a computer was first launched by Amazon in 2005; cleverly named Mechanical Turk for the 18th century human chess-maestro, who masqueraded as a machine, the site recruits people to perform ‘human intelligence tasks’ for small sums of money. Companies employ Turk for wide-ranging jobs, such as comparing two similar products or picking up visual cues from a video.

Prioritization, judgment and reasoning after all are the hallmarks of the human mind. It is no surprise then, that scientific research, which is increasingly being performed by computers, could never match the theoretical prowess of that nerdy scientist who sits at his bench for hours putting together an experiment. As Chris Anderson writes in Wired, while a computer saves a human brain the task of theoretizing, it does so by sidestepping the processes of analysis and causation. It churns up spools of data, and merely uses patterns and statistics to correlate. This task, while cumbersome, is still far easier than innovating scientific experiments to answer basic questions – something the brain has done for decades.

But it is the same haphazard human mind that dreams up a foolproof experiment that also believes in ghosts and goblins and aliens. So, ChaCha might occasionally tell you that being gay is a sin or give you the wrong schedule for a sports team, and sometimes, as Jason Pontin found, even simply throw up its hands in despair.

But that’s why we love it. It’s fallible, and it knows it. On a good day, it learns from its mistakes. On a bad day it screams in frustration.

My computer on the other hand, no matter how hard I hit it, just sits there gleaming.